RaivenTracks: Branching Provenance for Conversational Visualization Workflows
3.40T1 sourcearXiv cs.HC
Source record
Published by arXiv cs.HC (T1 source). The original is at https://arxiv.org/abs/2608.14869.
Pipeline notes
The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.
SummaryRaivenTracks extends the Raiven DSL visualization pipeline with branchable, persistent checkpoints to address statelessness in LLM-driven visualization systems. It uses a two-level state management architecture combining a version tree with undo/redo over runtime settings. A formative pilot with three visualization researchers shows initial usability and identifies design directions.
Why it mattersTargets the persistent problem of context loss when refining LLM-driven visualizations. The branchable-checkpoint pattern over DSL specifications is a concrete approach for scientist-in-the-loop AI oversight.
Cited by
No citations on record.
